* Disclaimer: When I say “math” I’m referring to the math that AI models rely on  (or technically, the math that the AI models ARE), in the first place. In other words, I’m referring to Calculus, Linear Algebra, Statistics/Probability, and higher level maths.

Everybody has a subject that they dislike in school. For some, it’s english: having to write one essay after another and read a large amount of books may lead to the disapproval of this course of study. For others, it’s science: maybe biology or chemistry did not connect with them. However, for many people, math becomes the bane of their existence. Whether it’s an inadequate teacher, the complexity of the subject, or just an unfavorable experience with the class, numerous people end up gaining a distaste for mathematics. This hatred for math becomes a problem for many reasons, one of them being the necessity of math in fields such as, you guessed, artificial intelligence. 

With the rise of AI in the past few years, many strive to join the industry in some way. Whereas this may seem like a good idea, many point to the fact that there are barriers of entry to the field. One such barrier is the supposed need to be an expert at math in order to succeed in AI. This is especially frightening to those who, as stated before, are not exactly the biggest math fans in the world. In this article, I want to go over whether you really need to be good at math to enter the field.

The Quick Answer

Getting to the point quickly, the answer to this question is dependent on many factors, such as which exact profession you are trying to take on. If you want to be an AI researcher who creates new architectures for neural networks or tries to find out how a deep learning model really “learns” (mechanistic interpretability), then YES, you will need to know a LOT of math. However, generally speaking, you DO NOT NEED TO KNOW MATH TO “HOP ON THE AI TRAIN”, but to an extent. Most frameworks that you will use when creating AI models abstract the mathematical aspects of AI, making it so that you can create an entire pipeline for creating models with only a high level understanding of what is really going on. Basically, you can solve real world problems with AI models that you create by only understanding the programming side of things while knowing minimal amounts of the mathematical side. Nonetheless, there are still caveats to this, so let’s discuss those next.

Profession

As said before, the level of math that you need to know and understand relies heavily on which exact profession you are trying to enter in the AI field. For example, most AI researchers will have to be good at math (although that in itself depends on WHAT they are researching). On the other hand, the average prompt engineer won’t really need to know Multivariable Calculus in their career. Other occupations like a Data Scientist or AI Engineer also don’t need to know math, but their case is a bit different. Even if prebuilt libraries do abstract away much of the complex math, it is still good to obtain at least a high level understanding of what is truly going on behind the scenes. If the extent of your knowledge doesn’t go beyond a bunch of syntax that you have memorized, then it shows that you have room for improvement. It is also important to note that even if you don’t need to have a full mathematical understanding of what is going on in the background, there is still no harm in learning some of the math behind things. In fact, this may come as a shocker, but learning will only help you, not harm you. With that in mind, here are a few resources out of many that you can use to learn math:

Learn the General Math needed for AI:

• Professor Leonard
• Khan Academy

Learn Math specifically in relation to AI:

• 3Blue1Brown
• StatQuest

Passion

Many people only want to join the AI field due to its large growth over the past few years and, more importantly, for the $$$. If your end goal is to just get a job in the field and make a living (which is fair, don’t get me wrong), then you don’t need to worry about a lack of math knowledge being a detriment to your career. On the other hand, if you are truly passionate about AI and want to learn as much as you can about it, then you will definitely need to deal with the math behind it eventually. 

To summarize, the necessity of knowledge in math is dependent on various factors, such as the exact profession you plan on joining in the AI field and your passion for AI. Generally speaking though, you do not NEED to know math or “be good” at math in order to have a successful career in the field. This does not take away from the fact that you should know; at least at a high level, about what is going on behind the scenes when you write and run code. And remember, knowing the math is still going to be extremely helpful for you. 

In 1920, a play called Rossum’s Universal Robots (R.U.R) was released to the public. R.U.R was the first play to depict the human race being completely annihilated and overthrown by robots, a plot premise that has been reused in multiple pieces of media ever since R.U.R’s introduction. Fast forwarding to the past few years, Artificial Intelligence has now exploded in popularity both as a field of study and a topic of conversation. This has culminated in “robots taking over” transitioning from a fictional narrative element to a potential event that humans will have to face in reality. This leads us to ask: is this fear of robots a truly dire and worrisome issue, or are people overreacting to the large surge that AI has had on the world? However, before answering this question, we need to understand what AGI is. 

AGI and it’s Ambiguity 

“AGI” is an acronym for “Artificial General Intelligence” and is meant to describe a type of AI model that is capable of completing a diverse set of tasks at around the same level that a human could. It can also be used to represent a stage or step in the overall process of the evolution of AI. While AGI has not been developed yet*, many still discuss its potential to benefit humans by improving fields like healthcare and solving more complex problems. However, many also see AGI as the beginning signs of a “robot takeover”, where AI starts to take jobs and replace humans in a variety of spaces. 

While AGI may seem like a simple concept at a glance, you might be surprised to realize its ambiguity when you try to define it yourself. The blunt truth is that it is hard to accurately pinpoint a definition or telltale sign of AGI. If you were to go up to 20 people and ask them to define AGI, there is a good chance that there will be noise in their answers. For example, some say that AGI is a system that can generally outperform humans on most tasks while some say that AGI is a system that is on par with humans on most tasks. On the other hand, there are also companies such as OpenAI that define AGI as an AI system that can generate $100 billion in profit for the company. Overall, what I am trying to get across is that AGI is a complicated stage in the evolution of AI systems and it is hard to tell when we will achieve AGI. It is still important to note that this does not take away from the importance of discussing AGI and its consequences, both positive and negative.

*Some argue that AGI has been achieved while some say that we still have some way to go. The difference in opinions partly stems from people having different definitions of AGI, as discussed in the above paragraph

How will AGI affect us?

Before, I only touched on the effects of AGI. Now, I want to go deeper into the implications of AGI

(Potential) Pros:

Revolutionizing Industries

AGI has the potential to transform industries like healthcare, education, and scientific research. Imagine a world where AGI diagnoses diseases more accurately than any human doctor, where AGI creates personalized treatment plans, or even develops cures for conditions that have eluded scientists for decades. In education, AGI could tailor learning experiences to individual students, making education more accessible and effective.

(Potential) Cons:

Job Displacement and Economic Inequality

One of the most immediate concerns is the displacement of jobs. While automation has always been a part of technological progress, AGI could replace entire professions that one might assume requires human innovation and creativity. This could drastically increase economic inequality, as those who control AGI systems gain wealth while others struggle to adapt.

Ethical and Safety Concerns

There’s also the risk of AGI being misused or behaving unpredictably. If AGI is developed without proper safeguards, it could lead to unintended consequences, such as AI systems making harmful decisions or being weaponized for malicious purposes. The question of whether AGI can act ethically—or even understand ethics—is one that remains unanswered

Will Robots take over?

The idea of robots “taking over” humanity usually goes hand in hand with a dystopian future seen in movies like “The Terminator” or “The Matrix”. While a robot takeover similar to the ones in these movies is unlikely, at least in the near future, they do show legitimate concerns about AGI surpassing human control. However, it’s important to remember that AGI, just like AI in general, is a tool whose impact depends largely on how we choose to develop and govern it.

Governments, tech companies, and researchers have to collaborate to ensure AGI is aligned with human values and priorities. We need to implement clear practices such as ethical guidelines, robust safety measures, and more transparent development procedures. By implanting such practices into society and institutions, we can effectively reduce the risks that AGI presents and use AGI as a tool for “good”. 

What can we do?

Overall, AGI holds immense promise, but it also comes with significant challenges. The threat of AGI is dependent on the decisions that we make and the actions that we take today. In general, I believe that AGI will not have as big of a negative impact as we might see in fictional media. However, it is still an important area of discussion. If we allow open conversations, invest in responsible AI research, and prepare for the societal shifts that AGI will bring, then the exploitation of AGI will be avoidable.

So, will robots take over humanity? Well, probably not. But that depends not on AI itself but how we choose to use it. 

.

Let’s get straight to the point: there is no one BEST way to “learn AI” and join the industry (yes, I click-baited you). Instead, you have many options, some being better than others, rather than there being a completely optimal one. This article aims to guide you through some options while giving you the pros and cons of each so that you can start your own journey in this increasingly important field. 

Replicate an AI project

To define what I mean by “replicate an AI project”, I’m specifically referring to finding a popular AI project online that has been done before and trying to replicate it by yourself with as little help as possible. Adding onto that last point, it is fine to use documentation for whatever framework or library you are using or use StackOverflow when you find yourself stuck. In fact, using these resources will most likely help you learn the skills that you will use when, in the future, you start creating your own AI projects. Just try to not follow a tutorial that explicitly gives you the exact steps and code to create a functioning final product, as that defeats the whole purpose of replicating the project in the first place (more on this in a bit). 

Pros: 

By replicating a project that already exists, you get hands-on experience with coding an AI project while not having to worry about large roadblocks during your work. You get an introduction to the general process of creating an AI project, you get to freely make mistakes and subsequently learn from them, and if you want to go deeper, you can extend the project by testing different models, trying to collect or curate more data, doing hyperparameter tuning, or any other ideas you can come up with. Another pro is that if you are ever in a pinch and don’t know where to go, there are tutorials and code notebooks online that can help redirect you to the right path. However, the important idea to keep note of is that you should only use these resources when you are REALLY stuck. Treat this project as a practice test with an answer key: if you immediately look at the answer key after barely trying to find the answer yourself, you’ll end up doing nothing but wasting your time and being no smarter than you were before. 

Cons: 

This option may be daunting as it requires you to get out of your comfort zone and try to do something completely new without any guidance tailored towards you. This can be amazing for some people, but not so great for others, especially if you have no programming experience or if you are not good at learning by yourself and need a teacher figure to help you out. Also, if you inevitably end up just following a tutorial to create the project, then you’ll end up just wasting your time as said before.

Overall, replicating an existing AI project is an amazing option. You have the freedom to make mistakes, don’t have to worry about roadblocks, and you get hands-on experience that you can apply to future endeavors. The only real drawback to this option is that it may not fit the needs of certain people (which is completely fine). 

Before moving onto the next option, I want to quickly give two examples of projects that are good to start off with:

1. House Pricing Dataset
   • Goal: Predict house pricing based on the given input
   • General Guideline: Use linear regression, then try to test other regression models
   • Good introduction to general Machine Learning (ML)
2. Fashion MNIST
   • Goal: Predict clothing classification based on given image
   • General Guideline: Start with logistic regression or SVM, then try out Neural Network
   • Good introduction to Deep Learning (DL)

There are also many sources such as kaggle and youtube that have AI projects that you can aim to replicate.

Follow a course

Another option is to find an introductory course to AI and use that course as your gateway into the field. This option is more straightforward than the previous option, so we can just hop straight into the pros and cons. 

Pros:

Assuming you choose a good course, following a course can be a great segue into the AI field. You get a large burst of information and get to be taught by an industry professional. There are usually hands-on projects where you code along with the instructor, which also allows you to gain the experience and skills that you need when you start working in the AI field for real. Expanding on that “large burst of information” I mentioned before, courses can give you a lot of knowledge on the AI field, specifically by examining different sub-fields in AI, looking at different frameworks and libraries that are useful, and introducing vast amounts of terminology that you otherwise may not have ever learned. 

Cons: 

The largest con of following a course lies in that first statement I made in the pros section. Basically, there is a good chance that the course you choose ends up being flawed or possibly even a scam. There are many problems that a course can have, such as a lack of hands-on activities, having an unqualified instructor, containing too much conceptual jargon, as well as others. Another large disadvantage with courses is that you are prone to forgetting a majority, if not everything, of what you learned, even with the more hands-on courses. This is especially true if you do not apply the skills you acquired after you take the courses. 

Overall, following an online course is a relatively good option, but has many problems of its own. However, I would not completely disregard online courses, as there are many instances where they can be very informative and add to your skills in the AI field. Many courses are also tailored to beginners, so they can be helpful when you are just starting in the field (which I’m assuming you are if you are reading this article). As long as you do your research on the course you take and apply what you learn afterwards, then this option is definitely worth considering. 

To end off this section, here are some courses I recommend, both free and paid. 

Free:

1. Machine Learning for Everybody– FreeCodeCamp.org
2. PyTorch for Deep Learning – FreeCodeCamp.org

Paid:

1. Complete Machine Learning and Data Science Course – Daniel Bourke
2. PyTorch for Deep Learning – Daniel Bourke (paid continuation of the other “PyTorch for Deep Learning” course listed in the free section)
   

* Note: Most courses, at least on Udemy, go on sale for a lot cheaper than they normally are really often. If you do buy a course, make sure to wait until the price is ACAP (as cheap as possible).  Also, I am not in any way affiliated with any of these courses.

To summarize, out of the two options, I would definitely recommend replicating a pre-existing AI project if you want an introduction to the field. It gives you hands-on experience and the skills that you need to succeed in real-world AI projects. And unlike with online courses, you will likely remember most of the skills and information that you gained through replicating the project. However, depending on how you prefer learning and overall personal preferences, taking a course could be a better option. As I said, there is no one best option to “learn AI” and join the field. The best thing to do is to just dive in and start learning. No matter how you start, you will end up succeeding as long as you stay consistent and work hard…most of the time. Anyways, thank you for reading, and I hope to see you again tomorrow!